同时用 Codex 和 Claude Code 的人，你们的工作流是怎么搭的？

I've only recently started paying seriously for AI tools, around when Astra came out, so I'm still figuring out what makes sense for me.

I've seen plenty of discussion about Claude offering better coding usage for the subscription price. At the same time, some Artificial Analysis comparisons seem to favour Astra on API cost per task. My current understanding is that these measure different things: API cost per task doesn't tell you how much usage a subscription actually includes. Is that a fair reading?

For me, Dot is becoming a fairly big reason to keep my OpenAI subscription. I've been using it across a few personal projects, and it helped set up a Notion hub with project summaries, next steps and things we're waiting on. Having that ongoing assistant alongside the coding tools feels useful.

I'm considering trying a month of Claude Pro for Claude Code as well. The idea would be to alternate Claude Code and Codex on the same local Git repository, keeping the Markdown documentation current so either can pick up the work. One could implement something and the other review it. I don't currently see much need for another coordination tool like Hermes or AionUI.

I can see myself getting quite attached to Dot, which makes switching subscriptions harder. I'd love to see it reach a cheaper plan eventually, though I know that's not promised.

Does this setup make sense to people using both? Any suggestions? Keen to hear what I'm missing.

TL;DR: Thinking of keeping OpenAI for Dot/Codex and trying Claude for extra coding capacity. Anyone doing this? What’s your setup, is paying for both worth it, and what would you suggest doing differently?